Analysis of Transcriptome, Mirnome and Genomic Profiles in Association with Clinical Outcome in a Prospective Series of Primary Plasma Cell Leukemia

Abstract

Abstract Abstract 3938 Primary plasma cell leukemia (PPCL) is a rare and very aggressive form of plasma cells dyscrasia, characterized by poorer outcome than multiple myeloma (MM). To provide insights into the biology of PPCL, we investigated 23 newly-diagnosed patients included in an open-label, multicenter, prospective, single arm, two-stage study aiming to explore efficacy and safety of lenalidomide and dexamethasone combination (LD) as first line therapy in previously untreated PPCL (median follow-up: 23 months; range 9–32). The primary endpoint of the study was the response rate, according to the criteria defined by International Myeloma Working Group, after 4-cycle therapy with LD over a 4-month schedule; among secondary endpoints were overall survival (OS) and eligibility to undergo autologous or allogeneic stem cells transplantation (SCT) after LD treatment. Herein, we took advantage of the FISH characterization and the microarray analysis of transcriptome, miRNome and copy number configurations of the PPCLs to investigate whether a correlation could exist between transcriptional features or allelic imbalances and clinical outcome. FISH was used to detect the main IGH translocations. The gene expression profiles of highly purified plasma cells from PPCLs cases were generated on GeneChip® Gene1.0 ST arrays. Expression values were normalized using robust multi-array average (RMA) procedure. MicroRNA profile were generated on Agilent Human miRNA Microarray V2. Expression values were extracted with Agilent Feature Extraction Software v10.1; quantile normalization was applied on raw data using R aroma.light package. GeneChip® Human Mapping 250K NspI arrays was used for genotyping. Copy number was estimated using circular binary segmentation and normalized on FISH data using R DNA.copy and FBN packages, respectively. To assess correlation between expression values and OS, R globaltest package was used to generate the linear regression model in which the distribution of the response variable is modeled as a function of the expression levels of each gene/miRNA. Our analysis indicated that all but three of the PPCLs had one among t(4;14) (13%), t(11;14) (39%) or MAF-associated translocation (35%). However, neither any of them nor any of the numerical alteration involving 1p, 6p, 8p, 13q, 14q, 16q, 17p (loss) and 1q (gain) as assessed by SNP-arrays were correlated with OS. As well, no correlation between response to treatment with LD and the prevalence of these cytogenetic alterations was evidenced. Of the 1145 most variable gene across the PPCL dataset and OS, 27 reached a highly significant correlation (P<.01) with OS. This 27-gene model was able to dissect the PPCL into two groups, one of which containing 6 cases with poorer outcome. In multivariate analysis, this model retained independency from all the cytogenetic alterations, as well as from age, sex, LDH levels, renal function and hematologic parameters. The 27-gene model was not independent of patients being subjected to autologous SCT, indicating that this therapeutic approach points definitively towards a more favorable outcome. Similarly, we assessed the relationship between each of the 114 most variable miRNAs across the dataset and OS. Two miRNAs reached a significant correlation (P<.01) with OS (miR-92a and miR-330–3p), allowing the division of samples into two groups with different outcome. In multivariate analysis, both retained independency from all the cytogenetic alterations [except del(8p) as regards miR-330–3p] and from other parameters, but not from autologous SCT. Finally, three genes (CYB5D2, EDEM3 and YIPF6) and four miRNAs (miR-497, miR-106b, miR-181a* and miR-181b) were identified having expression levels correlated with response to the first-line treatment with LD. Overall, this study represents the first integrated approach on a prospective study investigating genes and miRNAs expression and genotyping configuration in PPCL, indicating specific genes and miRNAs with relevance in the clinical outcome of the disease. Disclosures: No relevant conflicts of interest to declare.